Monocular Intra-vehicular Distance Estimation using Front Vehicle License Plate

Abstract

Intra-vehicular distance estimation is crucial for advanced driver assistance systems (ADAS), since it is capable of efficiently avoiding the dangers while driving. Most methods for distance estimation use the height of the camera above the road level, which is difficult to obtain. In this paper, a measurement method for license plate detection based on monocular distance estimation is proposed. The method consists of license plate localization, digital segmentation, and distance measurement modules. At first, license plate is detected using traditional method. To reduce the interference, a trapezoid area in front of the vehicle is chosen as the region of interest (ROI), and Sobel operator is employed to perform edge detection. Later, segmentation is performed followed by distance calculation using the heights of the digits on the license plate. The actual distance is then estimated by comparing the real height and the pixel height of the license plate digits that are obtained with a single camera. The experimental results show improved performance than the previous methods.